13.2 Proven Approach: Our Methodology Explained

{(x1, y1), (x2, y2), …, (xn, yn)}, where xi represents the image of the ith normal sample, and yi ∈ {0, 1} indicates its label. For all normal samples in SN, yi = 0, indicating that there are no defects in xi.

  • Patch Extraction: Each normal image xi is then divided into k smaller, overlapping patches pi = p1i, p2i, …, pki. The size and overlap of these patches are carefully chosen to balance detailed local analysis and computational efficiency. The patches from all images in SN construct a set P = {p11, p21, …, pk1, p12, p22, …, pk2, …, p1n, p2n, …, pkn}


  • Leave a Reply

    Your email address will not be published. Required fields are marked *